How dedinsaf/parentai works
ParentAI is a localhost homework-help application combining a browser avatar/camera with Python services. The main workflow captures a parent profile and voice sample, then listens for a child’s question, transcribes it, manages tutoring turns, and presents a spoken, lip-synced reply. The graph includes documented avatar construction, presence/listening, and voice options. Some implementation details are unsampled; those subsystem links are omitted or marked unverified rather than inferred from names alone.
Main entry points: app.js Lesson UI · server.py HTTP API · runtime.py Profile and local state · dialogue.py Dialogue sessions
Components 19
Browser experience
- Lesson UI
web/app.js - Scan and audio utilities
web/core.mjs - Speech gate
web/listening.mjs - Head and body geometry
web/head-geometry.mjs - Face texture atlas
web/face-atlas.mjs
Local application
- HTTP API
server.py - Application launcher
run_app.py - Profile and local state
runtime.py - Camera assets
download_assets.py
Tutoring and speech
- Dialogue sessions
dialogue.py - Homework tutor
tutor.py - Speech recognition
speech.py
Voice generation
- Voice profile generation
voice.py - XTTS engine
voice_engine.py - XTTS worker
voice_worker.py - Lip-sync timeline
lipsync.py
Other components
- Parent / child
- Yandex SpeechKit
- YandexGPT
Connections 21
- Parent / child → Lesson UI: uses
README.md - Lesson UI → HTTP API: calls API
README.md - Lesson UI → Scan and audio utilities: uses scan/audio
web/app.js - Lesson UI → Speech gate: uses speech gate
web/app.js - Lesson UI → Head and body geometry: builds avatar
web/app.js - Lesson UI → Face texture atlas: uses atlas
web/app.js - Application launcher → Profile and local state: creates runtime
run_app.py - Application launcher → HTTP API: starts server
run_app.py - HTTP API → Profile and local state: reads profile
server.py - HTTP API → Dialogue sessions: routes dialogue
server.py - HTTP API → Homework tutor: configures tutor
server.py - HTTP API → Speech recognition: configures speech
server.py - Profile and local state → Camera assets: starts asset setup
runtime.py - Profile and local state → Voice profile generation: starts voice setup
runtime.py - Profile and local state → XTTS engine: warms XTTS
runtime.py - Speech recognition → Yandex SpeechKit: transcribes audio
README.md - Dialogue sessions → Homework tutor: requests answer
dialogue.py - Homework tutor → YandexGPT: generates reply
README.md - Dialogue sessions → XTTS engine: synthesizes reply
dialogue.py - XTTS engine → XTTS worker: dispatches work
- Dialogue sessions → Lip-sync timeline: builds cues
dialogue.py